To followers of this presidential election, looking at the horse race can feel like a roller coaster. When viewed from certain angles and in certain perspectives, it looks like President Obama is a solid favorite to be reelected. Most public polls have him up, and he appears to be running ahead of his national numbers in the key swing states.

But viewed from other angles, Mitt Romney can seem stronger than these polls show. You can look at the changes in the overall political landscape since 2008 and conclude that Romney is being underestimated.

What is underlying these disparate views of the election is that no one is sure of what the electorate will look like this November.

2008 had the most Democratic electorate in a generation. Accordingly, Barack Obama was the first Democratic presidential candidate to earn over 50.5 percent of the popular vote since Lyndon Johnson in 1964. Thirty-nine percent of the 2008 electorate self-identified as Democrats, compared to 32 percent who identified as Republicans.

Most public polls are assuming a turnout model roughly similar to the 2008 electorate. If this was the case, Obama would certainly win. One could make the case that Democrats are more genuinely enthusiastic for Obama than Republicans are for Romney. And it does appear that Obama is running ahead of where a generic Democrat would be running.

There is another way to look at the electorate. What if the composition looked more like the 2004 or 2010 elections? In both of these elections, the Democrat/Republican self-identification was identical. In 2004, Kerry won independents by one point, but Bush was able to win by getting 11 percent of Democrats to vote for him, compared to 6 percent of Republicans voting for Kerry. 

In 2010, Republicans won a larger victory because they won independents by double digits. If this was true, a Romney win would be hiding in plain sight. Most signs do not point to a 2008 repeat in turnout disparity, but a top to bottom look at races across the country also doesn’t seem to point to a 2010-style Republican wave either.

Most of the uncertainty over where the state of the race is and will be on Election Day is over the partisan makeup of the electorate. Around 95 percent of those who identify with a party will vote for that party’s presidential candidate. Because this is such an ingrained preference, approximating the correct partisan breakdown of the electorate is the most critical part of getting an accurate view of the electorate. If pollsters presumed a 2008 turnout model but the actual election revealed a 2004 model, then there would be a large disparity between final predictions and the results.

Besides partisanship, the weighting of racial groups is also very important for determining the outcome. In 2008, 13 percent of the electorate was African-American—an outsized turnout compared to past elections. In 2004, African-Americans were 11 percent of the electorate. Because African-Americans are nearly unanimously in favor of President Obama, a one percent change plus-or-minus in the share of the electorate can make a great difference.

Hispanic turnout is also something to watch. Even though the Hispanic share of the population is increasing, the share of the electorate that is Hispanic has remained stuck at 8 to 9 percent since 2004. If Hispanic turnout stays the same or, as some reports suggest, declines then that can be a determining factor in a series of swing states like Florida, Colorado and Nevada.

Other subgroups can turn the election based on their relative share of the electorate. The 18 to 29-year-old vote was 1 percent greater in 2008 than in 2004. A reversion to past turnout levels would hurt President Obama. The proportion of the electorate that is married or single is worth watching, since marrieds tend to vote Republican and singles tend to vote Democratic. The religious composition is also of importance since nothing for white voters will predict political partisanship better than church attendance. Shifts among these subgroups are also to be looked at for hints of the final outcome.

No one can be sure of what the electorate will look like come November 6. The drama of this election is due to this above all else.

Chris Palko works as an assistant media analyst at Smart Media Group, a Republican political media buying agency in Alexandria, Va. He is a graduate of American University and George Washington University’s Graduate School of Political Management.

A version of this post was also published on Smart Media Group’s blog, Smart Blog.